Combining fragment docking with graph theory to improve ligand docking for homology model structures

Area of Science:

• Computational Chemistry
• Structural Biology
• Drug Discovery

Background:

• Protein-ligand docking accuracy is often limited by input receptor structure quality.
• Computationally predicted receptor structures pose significant challenges for reliable docking outcomes.

Purpose of the Study:

• To introduce and evaluate a fragment-based docking method designed to reduce sensitivity to receptor structure inaccuracies.
• To compare the performance of fragment-based docking against non-fragment approaches using both experimental and modeled receptor structures.

Main Methods:

• Docking small, rigid fragments individually to generate numerous poses within the receptor binding pocket.
• Employing a graph theory maximum clique algorithm to identify optimal combinations of fragment poses for complete ligand alignment.
• Determining complete ligand binding poses based on the identified fragment alignments.

Main Results:

• Achieved ligand poses with <1 Å root mean square deviation (RMSD) from experimental binding positions for Cytochrome P450 (CYP450) complexes.
• Recovered ligand poses with <3 Å RMSD for unbound docking using modeled structures, including SARS-CoV-2 protease.
• Demonstrated fragment-based docking's effectiveness with approximately modeled receptor structures.

Conclusions:

• Fragment-based docking offers improved reliability when utilizing computationally modeled protein structures.
• This method shows potential for enhancing drug discovery pipelines by overcoming limitations of receptor structure accuracy.